Vol. 174, No. 18JOURNAL OF BACrERIOLOGY, /Sept. 1992, p. 5941-59520021-9193/92/185941-12$02.00/0Copyright © 1992, American Society for Microbiology

A Rhizobium meliloti Lipopolysaccharide Mutant Altered inCompetitiveness for Nodulation of Alfalfa

A. LAGARES,l 2t* G. CAETANO-ANOLLES,3 K. NIEHAUS,4 J. LORENZEN,4H. D. LJUNGGREN,2 A. PUHLER,4 AND G. FAVELUKES'

Cdtedra de Quimica Biol6gica I, Facultad de Ciencias Exactas, Universidad Nacional de La Plata, 47y 115 1900-LaPlata, Argentina1; Department ofMicrobiology, Swedish University ofAgricultural Sciences, Uppsala, Sweden2;

Plant Molecular Genetics, Institute ofAgriculture and Centerfor Legume Research, The University ofTennessee, Knoxville, Tennessee 37901-10713; and Lehrstuhlfiir Genetik, Fakultat fur Biologie,

Universitat Bielefeld, Postfach 100131, D-4800-Bielefeld 1, Germany4

Received 16 April 1992/Accepted 21 July 1992

A transposon Tn5-induced mutant of Rhizobium meliloti Rm2011, designated Rm6963, showed a roughcolony morphology on rich and minimal media and an altered lipopolysaccharide (LPS). Major differencesfrom the wild-type LPS were observed in (i) hexose and 2-keto-3-deoxyoctonate elution profiles of crude phenolextracts chromatographed in Sepharose CL-4B, (ii) silver-stained sodium dodecyl sulfate (SDS)-polyacrylamidegel electrophoresis patterns of crude and purified LPS fractions, and (iii) immunoreactivities otherwise presentin purified LPS of the parental strain Rm2011. In addition, Rm6963 lost the ability to grow in Luria-Bertanimedium containing the hydrophobic compounds sodium deoxycholate or SDS and showed a decrease insurvival in TY medium supplemented with high calcium concentrations. The mutant also had altered symbioticproperties. Rm6963 formed nodules that fixed nitrogen but showed a delayed or even reduced ability tonodulate the primary root of alfalfa without showing changes in the position of nodule distribution profilesalong the roots. Furthermore, 2 to 3 weeks after inoculation, plants nodulated by Rm6963 were smaller thancontrol plants inoculated with wild-type bacteria in correlation with a transient decrease in nitrogen fixation.In most experiments, the plants recovered later by expressing a full nitrogen-fixing phenotype and developingan abnormally high number of small nodules in lateral roots after 1 month. Rm6963 was also deficient in theability to compete for nodulation. In coinoculation experiments with equal bacterial numbers of both mutantand wild-type rhizobia, only the parent was recovered from the uppermost root nodules. A strain ratio ofapproximately 100 to 1 favoring the mutant was necessary to obtain an equal ratio (1:1) of nodule occupancy.These results show that alterations in Rm6963 which include LPS changes lead to an altered symbioticphenotype during the association with alfalfa that affects the timing of nodule emergence, the progress ofnitrogen fixation, and the strain competitiveness for nodulation.

Bacteria of the genera Rhizobium and Bradyrhizobiumhave the ability to infect roots of leguminous plants to formnitrogen-fixing nodules (1, 43). The resulting symbiosis,highly specific with respect to the involved partners, devel-ops along an ordered multistage interaction, with participa-tion of plant and rhizobial components and specific solublesignals (26, 33). Thus, a complex, developmentally regulatedassociation takes place under the coordinated expression ofseveral symbiotic genes in the plant and the bacteria (44, 49).

Several bacterial surface components have been investi-gated and proved to be necessary for a compatible interac-tion of the bacteria with the host plant. Glucans (31),exopolysaccharides (5, 20, 21, 42, 45), and lipopolysaccha-rides (LPS) (6, 13, 19, 47) all appear to play a role duringsymbiotic interaction, especially during plant recognitionand infection (1, 11).The symbiotic role of LPS-and also exopolysaccha-

rides-was at first investigated by analyzing their interac-tions with legume lectins or roots (35, 38, 41). With thesekinds of studies, appropriate amounts of purified LPS fromdifferent rhizobia were shown to trigger metabolic events inwhite clover affecting infection thread development (24). By

* Corresponding author. Electronic mail address: antonio@biogenel.uni-bielefeld.de.

t Present address: Lehrstuhl fur Genetik, Fakultat fiir Biologie,Universitat Bielefeld, Bielefeld, Germany.

a different approach, in an attempt to characterize theparticipation of LPS in symbiosis, changes in the composi-tion of LPS were found among nodulating and nonnodulatingstrains of Rhizobium leguminosarum bv. trifolii (16, 50, 51)and Rhizobium fredii (17) after the loss of whole plasmids.Various mutants with alterations in LPS were defective inthe symbiotic association at different stages of development.Single-transposon TnS mutations in R. leguminosarum bv.phaseoli (19, 46), bv. viciae (25, 47), and bv. trifolii (6) andBradyrhizobium japonicum (56) altered simultaneously bothLPS structure and symbiosis. The kind of symbiotic alter-ations derived from these LPS changes varied with thebacteria examined. In R leguminosarum bv. phaseoli andbv. viciae, rough variants were clearly defective in infectionand nodule development (46, 47). It has been proposed thatR leguminosarum LPS could determine surface charge orhydrophobicity, both important for bacterial release fromthe infection threads (25). However, the expression of aheterologous 0 antigen in R. leguminosarum bv. trifolii canstill support a normal symbiosis (6). Despite this apparentlyrelaxed requirement for a symbiotically active LPS, changesin specific epitopes of the 0 antigen occur during the processof bacteroid differentiation (32, 54). In R leguminosarumbv. phaseoli, similar changes also occur in free-living bacte-ria grown at low pH and were shown to be independent ofthe presence of Sym plasmid genes (58). Progress in under-standing the chemistry of LPS is currently being made to

5941

5942 LAGARES ET AL.

TABLE 1. Microorganisms and plasmids

Strain, phage, Relevant characteristic(s) Source or referenceor plasmid

StrainsR. melilotiRmL5-30 Wild type G. Martinez, Dretz, UruguayRmlO21 Wild type S. R. Long, United StatesRm2011 Wild type, Smr J. Denarie, FranceRm6963 TnS mutant derivative of Rm2011, Nod' Fix', altered LPS, altered This work

in competitiveness for nodulation of alfalfa, Smr NmrE. coli

S17.1 MM294, RP4-2-Tc::Mu-Km::Tn7 integrated in the chromosome, Tpr 47Phage4M12 Transducing phage 29

PlasmidpSUP1021 TnS derivative of pSUP102, Ampr Cmr Tcr Nmr 47

account for this and other LPS structures that supportepitopes of symbiotic relevance (for a review, see reference14).The participation of LPS in the association between

Rhizobium meliloti and alfalfa is currently also under activeinvestigation. Although an early role for LPS in plant recog-nition during bacterial adsorption to roots has been sug-gested elsewhere (41), other results have supported thehypothesis of participation of LPS later in symbiosis (48, 62,63). Interestingly, mutations in some exo loci causing inef-fective nodulation were suppressed by lpsZ, a gene presentin Rm41-derivative strains, restoring the production of nitro-gen-fixing nodules (63). However, mutational alteration ofR.meliloti LPS appears to have little or no effect in symbiosis.This theory could be upheld by the evidence that from abroad group of LPS mutants, none appeared to be deficientin the association with alfalfa (23). All of these mutants wereable to fix nitrogen like the wild-type strain, and only someof them under a bacterial background carrying other surfacemutations resulted in Fix-. It was suggested, however, thatthis phenotype was due to a cumulative effect of severalindependent mutations (23).

In the present paper, we characterize changes in thebiochemistry and physiology of an R meliloti LPS mutantthat was able to fix atmospheric nitrogen but was clearlyaltered in symbiosis. We show that alterations in the mutantaffected the timing of nitrogen fixation and the ability tocompete for nodulation. The possible role of LPS during theassociation of the bacteria with alfalfa will be discussed.(A preliminary summary of this work was presented at the

13th North American Symbiotic Nitrogen Fixation Confer-ence, Banff, Alberta, Canada, 25 to 30 August 1991 [abstractP14p].)

MATERIALS AND METHODSPlants, microorganisms, and plasmids. Alfalfa (Medicago

sativa L. cv. Vernal) seeds were surface sterilized as previ-ously described (8). Briefly, alfalfa seeds were pretreatedwith 97% ethanol for 1 h and then treated for 10 min with0.2% (wt/vol) HgCl2-0.5% (vol/vol) HCl and four washeswith sterile water. The seeds were germinated on invertedwater-agar petri dishes and used in plant nodulation experi-ments. Microorganisms and plasmids are described in Table1.Media and growth conditions. Bacterial strains were cul-

tured in yeast extract-mannitol (YEM) medium (in grams perliter: yeast extract, 0.4; mannitol, 10; NaCl, 0.1; MgSO4.

7H21, 0.2; K2HPO4, 0.5 [pH 7.0]) for nodulation assays andin defined glutamate-mannitol (GM) medium (in grams perliter: sodium glutamate, 1.1; mannitol, 10; K2HPO4, 0.23;MgSO4 7H20, 0.1; and CaCl2, 0.005; trace elements andvitamins [in micrograms per milliliter]: H3BO3, 145; FeSO4.7H20, 125; CoS04. 7H20, 70; CUSO4- 7H20, 5; MnC12.7H20, 4; ZnSO4. 7H20, 108; Na2MoO4, 125; nitrilotriace-tate, 7,000; riboflavin, 20; nicotinic acid, 20; biotin, 20;HCl-thiamine, 20; HCl-pyridoxin, 20; calcium pantothenate,20; inositol, 120 [pH 7.0]) for LPS preparations. Bacteria ingenetic manipulations were grown in tryptone-yeast extract(TY) medium (3) or Vincent minimal medium or modifiedLuria-Bertani (LB) medium (containing, in grams per liter:glucose, 1; tryptone, 10; yeast extract, 5; and NaCl, 5 [pH6.80 to 6.85]) (60) as indicated. When appropriate, the mediawere supplemented with streptomycin (400 ,ug/ml) and/orneomycin (120 ,g/ml).

General TnS mutagenesis. The Escherichia coli strainS17.1 carrying the tra gene region of RP4 integrated in thechromosome was used to transfer the suicide vectorpSUP2021 carrying transposon TnS to strain Rm2011 (53).Neomycin-resistant (Nmr) transconjugants were isolated onTY medium containing the antibiotic.

Generalized transduction. Transduction of Nmr markerfrom Rm6963 to Rm2011 was done by using the transducingphage 4M12 according to the protocol described by Finan etal. (29).LPS preparations. Bacterial cultures of Rm2011 and

Rm6963 were grown in liquid GM medium to late log phase(optical density at 500 nm [OD500] = 0.5 to 0.8) and centri-fuged for 20 min at 10,000 x g. LPS from harvested cells wasextracted with a 45% phenol-water mixture as describedelsewhere (61). The aqueous phases after two extractionswere pooled and exhaustively dialyzed (membrane cutoff, 12kDa) against distilled water at 4°C. This crude LPS solutionwas lyophilized and stored at -20°C.LPS purification by chromatography on Sepharose CL-4B.

Crude LPS extracts were resuspended in 100 mM EDTA-triethylamine (pH 7.0) and the suspension was kept over-night at 4°C to enhance dissociation. LPS samples (6 to 12mg as hexoses in 1 to 2 ml) were separated in a SepharoseCL-4B column (2 by 60 cm) with 10 mM EDTA-triethyl-amine running buffer (pH 7.0) at 4°C (15).

Determination of hexoses and KDO. Crude LPS and sub-fractions obtained from the Sepharose CL-4B column wereassayed for hexoses by the phenol-sulfuric method (27). The

J. BACTERIOL.

V MELILOTI LPS MUTANT ALTERED IN SYMBIOSIS 5943

2-keto-3-deoxyoctonate (KDO) content was determined bythe method described by Karkhanis et al. (37).

Colorimetric measurement of hydrophobicity. The spectralshift of the lipophilic probe merocyanin 540 (64) in thepresence of hydrophobic compounds was used to estimatehydrophobicity. We observed a spectral shift to longerwavelengths when the cyanine reacted with solutions con-taining LPS. The coefficient R [R = (OD567/OD5OO)sample -(OD567/OD500)buffer] was linear with LPS concentrations inthe range of 20 to 400 ,ug/ml. A 900-,ul volume of each samplewas mixed with 100 ,ul of 50 ,ug of merocyanin per mlprepared immediately before use. After mixing, opticaldensities were measured and R was calculated. This methodwas used to analyze fractions corresponding to the maxi-mum of each peak eluted through the Sepharose CL-4Bcolumn.

SDS-polyacrylamide gel electrophoresis (PAGE). LPS prep-arations were solubilized in sample buffer (final concentra-tions: Tris-HCl, 0.0625 M; sodium dodecyl sulfate [SDS],2% [wt/vol]; ,B-mercaptoethanol, 5% [vol/vol], bromophenolblue, 0.001% [wt/vol]; glycerol, 10% [vol/vol]) and heated for2 min at 100°C. A 4% stacking gel and a 12% separation gelwith the Laemmli (40) buffer systems were used. Gels werefixed and silver stained as described by Tsai and Frasch (59).

Sensitivity of R. meot to hydrophobic compounds. Theability of R meliloti to grow in the presence of sodiumdeoxycholate (DOC) was tested by plating appropriate bac-terial dilutions in LB solid medium containing from 0.05 to 2g ofDOC per liter. In each case, the pH was adjusted to 6.80to 6.85. Viability was expressed as the percentage ratio ofcolonies developed in the presence and absence of DOC. LBmedium was also used to study the sensitivity of the mutantto SDS (0.1 g/liter).

Nodulation profiles in alfalfa primary-root and split-rootexperiments. Two-day-old seedlings were transferred to eth-ylene oxide-sterilized plastic growth pouches containing 10ml of nitrogen-free Jensen mineral solution (34). Three dayslater, primary roots were inoculated by dripping 100 pl ofbacterial suspension onto the root from the tip toward thebase. The positions of the root tips and smallest emergentroot hairs were marked on the plastic pouches immediatelyafter inoculation with the aid of a dissecting microscope at amagnification of x 12. The plants were cultured in a growthchamber at 80% relative humidity, at 26°C during the dayand 24°C during the night, for a photoperiod of 16 h, and ata photosynthetically active radiation rate of 250 mmol - s-1m-2. The number and relative location of individual noduleswere determined with a computer-linked graphic tablet (4).Split-root experiments were carried out as previously de-scribed (7, 10).

Coinoculation experiments. Alfalfa plants grown in 250-mipots filled with sand and nitrogen-free Jensen solution wereinoculated 5 days after germination with late-log-phase cul-tures of Rm6963 and Rm2011. To prepare the inoculants,both strains were mixed in different proportions to obtainbacterial concentration ratios that ranged from approxi-mately 107:107 to 107:1 bacteria . ml-1 (Rm6963 to Rm2011).Pots containing 30 plants were inoculated with 3 ml of thebacterial suspensions. In nodule occupancy studies, theuppermost root nodules were excised, surface sterilized in20 vol of H202 (15 min), and crushed in a 100-pl Fahraeus(28) mineral solution. The resulting suspensions were pickedin solid medium with 1 g ofDOC per liter of LB or 120 p,g ofneomycin per ml of YEM. The ability to grow in both mediawas considered evidence of dual occupancy or due to apossible reversion of the mutant. Competitiveness for nod-

ulation was assessed by the proportional representation ofeach strain in the nodules relative to its proportional repre-sentation in the inoculum (2).LPS immunoreactivity. Antiserum preparation and Ig pu-

rification. Antiserum to RmlO21 late-log-phase cells wasprepared as described by Somasegaran and Hoben (55).Immunoglobulins (Igs) from the crude serum were purifiedby ammonium sulfate precipitation and DEAE-cellulosechromatography as previously described by Clark and Ad-ams (22). Igs were monitored during the purification proce-dure by cellulose acetate electrophoresis.

(i) ELISA. The ability of Rm2011 LPS and Rm6963 LPS tobind homologous anti-R. meliloti Igs was estimated by anenzyme-linked immunosorbent assay (ELISA). Polystyrenemultiwell plates were incubated for 1 h at 37°C and overnightat 4°C with 100 ,ul of selected fractions eluted from theSepharose CL-4B column per well. The plates were thenpretreated for 1 h at 37°C with 100 pl of 3% skim milk perwell in phosphate-buffered saline (PBS) to prevent nonspe-cific binding of the antiserum to the plate. After washing, 100pl of anti-R meliloti 1021 rabbit gamma globulin (dilution,1:100) per well was added, the solution was incubated for 1h at 37°C and washed, and 100 pl of goat anti-rabbit IgGconjugated to horseradish peroxidase (1:1,000 [Sigma]) perwell was added and incubated for 1 h at 37°C. Antibodieswere always diluted in 3% skim milk-PBS. The substrate forcolor development was a solution of 2 mg of o-phenylenedi-amine (Sigma) per ml in 0.1 M Na2HPO4-0.1 M sodiumcitrate-100 pl of H202 (30%) (pH 5) per liter. The reactionwas initiated by the addition of 100 pl of freshly preparedsubstrate to each sample, which was incubated for 5 to 15min at room temperature, and stopped with the addition of100 pl of 4 N H2SO4. The A490 of the o-dinitrobenzeneproduct in the supernatant fluid was measured against abuffer blank which had been subjected to the entire ELISAprotocol.

(ii) Immunoblotting. For detection of LPS subfractionsthat supported a positive reaction in ELISA, LPS sampleswere electrophoretically fractionated by SDS-PAGE, trans-ferred to nitrocellulose, and analyzed by immunoblotting asdescribed by Strum et al. (57). The reactivity of eachcomponent was deduced by comparison of the immunoblot-ting with the silver-stained pattern of a duplicate gel.

RESULTS

Isolation of the mutant Rm6963. Mutants of Rm2011 wereobtained by general TnS mutagenesis with plasmid pSUP1011. About 10,000 neomycin-resistant transconjugants weretransferred to TY medium and Vincent minimal medium.Among the TnS mutants isolated, 91 showed an alteredcolony morphology. One of these mutants, Rm6963, pro-duced rough colonies. The mutant was more sensitive tohydrophobic conditions when tested on LB medium contain-ing appropriate amounts of DOC (1 g/liter) or SDS (0.1g/liter). Colony morphology and sensitivity to DOC and/orSDS suggested the presence of an altered LPS in Rm6963.General transduction with phage M12 showed that neomy-cin resistance was always cotransducible with sensitivity tohydrophobic compounds. This indicated that TnS was re-sponsible for the mutant phenotype.

Analysis of crude hot phenol-water extracts from Rm2011and Rm6963 on Sepharose CL-4B. Hot phenol-water extractsobtained from strains Rm2011 and Rm6963 were separatelychromatographed in Sepharose CL-4B, and the collectedfractions were assayed for hexose and KDO (Fig. 1). Three

VOL. 174, 1992

5944 LAGARES ET AL.

1200

800

4

z0

(I)

illI-(I)0

xU-i

400

0

1200

800

400

0

0 0.5

Kd

500

400

300

200

100

0 ;500 >

0-Hi0z

400 %

(0

300 °

200

100

FIG. 1. Sepharose CL-4B chromatograms of hot phenol extracts of strains Rm2011 (A) and Rm6963 (B). Each sample, containing 6 to 12mg as hexoses, was applied to the column in a volume of approximately 1.5 ml and run at 4°C (column diameter, 2 cm; height, 60 cm). Theflow rate was ca. 8 ml/h. Fractions of 4 ml each were collected and analyzed for the presence of hexose (0) and KDO (0) as described inMaterials and Methods. The void volume (V7) and included volume (Vi) for calculation of distribution coefficients Kd [Kd = (Vsample - V0)/Vj]were measured by using 1.5 ml of a mixture of blue dextran (0.05 g/ml) and CaCl2 (0.5 g/ml). V0, 66 ml; Vi, 136 ml.

peaks containing hexose were eluted, of which peaks PI(Kdma = 0.1) and PII (Kdmax = 0.5) contained KDO. Thepresence of KDO in peak PIII from both strains appeared tobe due to the tailing of peak PII. The results suggest that inthis chromatographic system, the LPS of Rm2011 was dis-tributed in both PI and PII (similar profiles were alsoobserved for crude LPS of the homologous strain RmL5-30,indicating that this pattern may be common to other wild-

type R meliloti strains). However, the hot phenol-waterextracts from Rm6963 and Rm2011 showed different behav-iors. In the mutant, the KDO content of peak PI (nearly 60%of the total) was considerably more than that of Rm2011 (5%of the total) (Table 2). This finding suggests an LPS altera-tion in Rm6963 detected as an increase in the KDO-contain-ing material eluting near the void volume of the column. Theshift in the elution of KDO corresponded to an increase in

TABLE 2. Analysis of hexose, KDO, and hydrophobic properties of peaks PI, PII, and PIII after chromatographyof bacterial crude phenol extracts in Sepharose CL-4B

,ug/peak (%)'LPS origin PI (Kd = 00 34)b for: PII (Kd = 0.35-0.63)b for: PIII (Kd = 0.64-1)b for:

Hexoses KDO Rc Hexoses KDO Rc Hexoses KDO Rc

Rm2011 918 (9) 156 (5) 0 1,832 (18) 2,310 (75) 1.4 7,624 (73) 626 (20) 0Rm6963 1,128 (19) 1,227 (59) 1.2 787 (13) 617 (29) 0.2 4,036 (68) 253 (12) 0a Numbers in parentheses correspond to percentages of the total seeded material.b For definition of Kd, see Results.c The hydrophobicity of each sample was estimated with the R coefficient by using the optical probe merocyanin 540 as follows: R = (OD567/OD500)sampIe -

(OD567/OD5O)buffer. Optical densities were measured in the presence of merocyanin (final concentration, 5 ,ug/ml). R values were obtained for fractions at themaximum of each peak.

J. BA=rRIOL.

1

R MELILOTI LPS MUTANT ALTERED IN SYMBIOSIS 5945

A B

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

1 5

{LPS 11 -ILLPS Ila M

FIG. 2. SDS-PAGE (A and C) and immunoblotting (B) of chro-matographed hot phenol-water extracts from strains Rm2011 andRm6963. Slabs from panels A and B (7 by 8 cm) were run in aMini-Protean II 2-D cell (Bio-Rad). Immunoblotting of panel B wascarried out with a gel duplicating the samples from the one shown inpanel A and with purified Ig from a rabbit antiserum against Rmeliloti RmlO21 as first antibody. The well samples were as follows:lane 1, Rm2011 crude hot phenol-water extract; lane 2, Rm2011 PI;lane 3, Rm2011 PII; lane 4, Rm2011 PIII; lane 5, Rm6963 crude hotphenol-water extract; lane 6, Rm6963 PI; lane 7, Rm6963 Pll; lane 8,Rm6963 PIII. Each well contained 1 pg (as hexoses) of the corre-sponding fraction. Panel C shows details of bands LPS II and LPSIla after being run in a gel (12 by 13 cm) to improve resolution (4p1g/well). The positions of the bands are indicated in the left margin.

the R hydrophobicity coefficient of peak PI (Table 2). Thehigh absorption of the optical probe merocyanin 540 at 567nm might be attributed to the possible presence of lipid A, apotential enhancer of the basal hydrophobicity of the bufferEDTA-triethylamine. A high R coefficient was measured inpeak Pll of strain Rm2011, correlating again with the pres-ence of KDO. No differences between both strains wereevident in peak PIII.SDS-PAGE analysis. Crude and chromatographically puri-

fied LPS extracts were analyzed by SDS-PAGE. Figure 2Ashows the silver-stained banding pattern of crude hot phe-nol-water extracts from Rm2011 and Rm6963 (lanes 1 and 5,respectively). According to criteria already applied to otherrhizobial LPS (6, 18), components in the gel were designatedLPS I or LPS II according to their electrophoretic mobilities.

PI

_EC00)'1.0

2.5

2

1.5

1

2.5

2

1.5

Marked differences among the silver-stained patterns ofparent and mutant strains were observed. The mutant LPSlost the band corresponding to component LPS I which ispresent in the parental LPS but showed a new component(marked LPS Ia) which migrated faster to the anode. Differ-ences were also evident in LPS II. A broad band markedLPS IIa had a faster mobility than its apparently homologouswild-type component LPS II. As shown in Fig. 2C, LPS IIand LPS IIa components were clearly resolved with longslab gels. Assuming that these components correspond tolipid A and core, the mutant should be affected in at least oneof these LPS structural regions. The increase in electropho-retic mobilities in both LPS I and LPS II suggests a decreasein the length or an increase in the charge of mutant LPSmolecules. Besides these changes, other components foundin the LPS of Rm2011 were located between LPS I and LPSII (marked LPS I/II) and were not observed at the sameposition in Rm6963 LPS. In our experimental conditions,LPS components did not resolve in clusters of several bandsor exhibit patterns of regular periodicity characteristic of theLPS heterogeneity described with other rhizobia (12).Components of the isolated peaks PI, PII, and PIII were

also analyzed by SDS-PAGE (Fig. 2A, lanes 2 to 4 and 6 to8). As expected from the KDO levels of the fractions (Fig.1), silver staining confirmed that most of the LPS fromRm2011 corresponded to peak PII, while LPS from Rm6963distributed between both peaks PI and PII (all samples wereloaded in amounts with equivalent hexose contents). Sincethe substrate for the LPS silver staining appears to be mainlythe lipid A (39), a coincidence between the best-stainedfractions and those with a higherR coefficient was expected.LPS immunocharacterization. The immunocharacteriza-

tion of LPS was carried out by immunoblotting and ELISA.Fractions corresponding to the maxima of peaks PI, PII, andPIII were tested in their reactivities against purified anti-R.meliloti Ig. Although the reactivities of peaks PI and Pllfrom Rm2011 correlated with the respective KDO contents(Fig. 2B, lanes 10 and 11), mutant LPS samples werecompletely unreactive (Fig. 2B, lanes 13 to 16). This couldbe confirmed by ELISA (Fig. 3). Under our experimentalconditions, the ELISA was highly sensitive, detecting in thecase of peak PII from Rm2011 up to 20 to 100 pg/well (ashexoses). The absence of reaction in LPS samples ofRm6963 could be confirmed at this level. Immunoreactivitywas strictly dependent on the presence of wild-type LPS.

PillPll2.5 T

2

1.5-

10.

0.5

'-Cooe Nooeo

_ED, CQt

- D

n-S _ _ _ _ _ _ _

dilution factorFIG. 3. Reactivity in ELISA of peaks PI, PII, and PIII from strains Rm2011 and Rm6963 against purified Igs from an anti-R. meliloti

antiserum. Fractions corresponding to the maximum of each peak (Fig. 1) were diluted as indicated and used as antigens. The ELISA was

carried out as described in Materials and Methods.

VOL. 174, 1992

5946 LAGARES ET AL.

C.)00

0

>% C

00

C-,

.

>0

atC.)

160

140

120

1004

80

60

40

20

0

0.1

140

120

1004

80

60

40

20

0

DOC (g/1) added to LB medium

0 20 40 60 80 100

[Ca++] (mM) added to TY mediumFIG. 4. Dependence of the viability of strains Rm2011 (0) and

Rm6963 (0) on the presence of increasing concentrations of DOC(A) or Ca2` (B). An appropriate dilution of bacterial cultures in thelog phase of growth containing 102 bacteria was plated in LBagar-DOC and TY agar-Ca2", and the number of bacterial colonieswas determined 4 to 7 days later. The results are expressed as

percentages of colonies for a given DOC or Ca2+ concentration withrespect to the control medium without any addition.

The total loss of reactivity in the mutant (all bands) indicatedthat LPS of Rm6963 had lost antigenic structures present inthe wide range of molecular forms of Rm2011 LPS (LPS I,LPS II, and LPS VII).The hexose-rich and KDO-poor material of peak PIII

present in both crude LPS preparations was neither silverstained nor recognized by the homologous Ig (Fig. 2B, lanes12 and 16). As shown in Table 2, peak PIII did not show thehydrophobic character detected in peak PI of the mutant andpeak PII of the wild type with the optical probe merocyanin540, suggesting the absence of LPS in this fraction. In otherrhizobia, material eluting at the position of PIII is often,B-(1-2)-glucan.

Sensitivity to hydrophobic compounds. The growth of mu-tant and wild-type bacteria in the presence of DOC (or SDS)was used to characterize the outer membrane as a barrier tohydrophobic compounds (23, 52). The mutant was moresensitive to increases in hydrophobicity of the medium thanthe wild type (Fig. 4A). When LB medium (pH 6.80 to 6.85)was supplemented with concentrations of DOC of 1 g/liter orhigher, only Rm2011 was able to grow. The mutant exhibiteda survival of 50% at about 0.7 g of DOC per liter. Apparentsurvival rates with values higher than 100% at very low DOCconcentrations were attributed to disruption of cell aggre-

gates caused by mechanic spreading during plating. Sensi-tivity to DOC was extremely dependent on the pH. At pHshigher than 7.0 (i.e., 7.4 to 7.5), the LPS mutant could grow(though deficiently) even in the presence of 1 to 2 g of DOCper liter.The addition of 0.1 g of SDS per liter to the LB medium

also clearly inhibited the growth of Rm6963 and to someextent the growth of Rm2011 (data not shown). Both DOCand SDS could be used for counterselection of the wild-typestrain against the mutant.

Sensitivity to high calcium concentrations. Strain Rm6963showed a decreased resistance to growth in TY medium withhigh calcium concentrations (Fig. 4B). Increases in theCaC12 concentration diminished the rate of bacterial growth.Negligible growth occurred at 100 mM, with colonies devel-oping only after 2 weeks. This alteration in the cell growthphysiology may be derived from the altered LPS, as reportedfor some LPS mutants of R. leguminosarum bv. viciae (36).

Nodulation of R. meliloti 6963. Nodulation assays in glasstubes containing nitrogen-free Jensen medium showed thatthe mutant produced apparently normal nitrogen-fixing nod-ules able to support growth of the inoculated plants. How-ever, when plants were grown in plastic pouches and werecarefully analyzed, nodules induced by Rm6963 developedthe characteristic pink coloration of normal indeterminatenodules several days after the appearance of nodules formedby wild-type bacteria. Plants inoculated with the mutantwere smaller and yellowed during the first 2 to 4 weeks ofgrowth. These plants showed a marked reduction in the freshweight of their green tissues (ca. 30%) 3 weeks after inocu-lation. This symbiotic phenotype paralleled the progress ofnitrogenase expression. Acetylene reduction assays showedthat, for example, at 20 days postinoculation, the rate offixation was 35 and 45 nmol of C2H2. plant-' h-1 forplants inoculated with the mutant and wild type, respec-tively. However, plants recovered later in their developmentwith an appearance that became comparable to that of thecontrol inoculated with Rm2011.Time course of nodulation. We studied the ability of parent

and mutant strains to nodulate alfalfa on primary and lateralroots (Fig. 5). Although there were no appreciable differ-ences in the speed with which nodules emerged on primaryroots (slope in Fig. 5 graph), nodules induced by Rm6963started to appear 4 days later than wild-type nodules (Fig.SA). In contrast, the time rate of nodule emergence on lateralroots was consistently lower for plants inoculated withwild-type bacteria (Fig. SB). After about 12 days postinocu-lation, nodules induced by Rm6963 in lateral roots increaseddrastically and reached levels that were higher than thoseobtained with Rm2011. At 1 month postinoculation, thenumber of nodules induced by the mutant on primary rootsranged between 40 and 100% of the number of wild-typenodules; on secondary roots, however, nodulation was up tofivefold higher (Fig. SB).

Distribution of nodules on primary roots. Figure 6 showsthe distribution of nodules on primary roots 9 days postin-oculation. Both strains were able to nodulate essentially thesame region of the roots. Since nodules induced by Rm6963emerged later, total nodulation on the primary root duringthe first week postinoculation was decreased, and this wasreflected by the number of nodules formed at differentlocations on the primary root (Fig. 6). In addition, we haveobserved that later than 15 days after inoculation, mutantand wild-type nodule distribution profiles were comparable.The observation that alfalfa primary roots were nodulated by

0oe B

aw @ S-.~~

III\

J. BA=rRIOL.

R. MELILOTI LPS MUTANT ALTERED IN SYMBIOSIS 5947

lU

8 A. Primary root

6

4o0

2

,o .,.. , ...,,......4 8 12 4 8 12 16 20 24

DAYS AFTER INOCULATIONFIG. 5. Kinetics of nodulation of alfalfa roots by strain Rm2011 and its LPS mutant, Rm6963. Sets of 84 and 73 plants were inoculated with

1.5 x 106 bacteria per plant of strain Rm2011 (0) and Rm6963 (0), respectively. Nodules on primary (A) and secondary (B) roots wereperiodically scored at regular intervals of time. Results are given as average numbers of nodules per plant and are representative of threeexperiments. Nodulation of the mutant was more pronounced in this experiment than in the other two.

both strains in the same regions suggests that strain Rm6963behaves normally during preinfection.

Split-root experiments. We examined the ability of Rm6963to elicit systemic feedback control of nodulation in alfalfa. Insplit-root assays, prior inoculation of one side of a split-root

1.5

z

ID

(/')LiJ

ID

0

z7

1.0

0.5

0.0

0.5

0.0, ,_.---I\

EH RT -5

system with Rm6963 suppressed nodulation of a wild-typebacterium on the other side by 44% (Table 3). Prior inocu-lation with Rm2011 resulted in stronger inhibition of nodu-lation (93%). Results show that mutant Rm6963 had adecreased ability to autoregulate nodule number, probably

B. Rm6963

*., 0

-10 -20

RELATIVE DISTANCE UNITS (RDU)FIG. 6. Distribution profiles of nodules generated on alfalfa roots inoculated with Rm2011 (A) or with its LPS mutant, Rm6963 (B). Sets

of 80 and 72 plants were inoculated with strain Rm2011 (2.5 x 106 0.2 x 106 bacteria per plant) and strain Rm6963 (1.4 x 106 + 0.3 x 106bacteria per plant), respectively. Profiles correspond to an early-nodule scoring 9 days after inoculation. The relative distance of each nodulewas expressed as a percentage of the distance of root tips (RT) to the smallest emergent root hairs (EH) determined for each plant. Theaverage distance of root tips to emergent root hairs was 2.9 1.2 mm for this experiment. Since the average rate of root elongation was 0.5+ 0.004 mm/h, 1 relative distance unit (RDU) is equivalent to approximately 6 h of root growth. The direction of root growth is from left toright. The results are from a representative experiment among three.

H-z

CLLuJ1-J1=f):DC)0z

0

0

B. Lateral roots o/0

0

/0

pp ~ ~ O'-

16 20 24 0

VOL. 174, 1992

4 f%

5948 LAGARES ET AL.

TABLE 3. Effect of prior inoculation with Rm2011 or its LPS mutant, Rm6963, on one side of a split-root system,on nodulation on the second side by an indicator straina

Location and inoculum Avg no. of nodules/lateral root for: Relative nodulationtype for side A Side Ab Side BC in side B

Above root tipsSham 032 2.23 (1.63-2.92)29 100Rm2011 (parental) 2.44 (1.94-3.01)3° 0.15 (0.04-0.35)33 7Rm6963 (LPS mutant) 0.43 (0.25-0.65)37 1.25 (0.80-1.78)28 56

Whole lateral rootSham 0.02 (0-0.14 ) 4.01 (3.12-5.01) 100Rm2011 (parental) 3.88 (3.02-4.85) 0.39 (0.16-0.71) 10Rm6963 (LPS mutant) 1.34 (0.90-1.87) 2.05 (1.42-2.80) 51a Split roots were inoculated at the time of marking root tip 1 on side A with Hoagland mineral solution (sham), with 3.7 x 106 bacteria of strain Rm6963 per

root, or with 8.6 x 106 bacteria of strain Rm2011 per root as indicated, and then inoculated 24 h later (root tip 2) on the opposite side (side B) with 7.2 x 106 bacteriaof the indicator strain RmlO21 per root. The average distance between root tip 1 and root tip 2 was 8.3 ± 1.5 mm. Mean numbers of nodules are based on thesquare root transformation, with confidence limits of 95% in brackets. Superscripts are the numbers of lateral roots per side examined.

b Above root tip 1.c Above root tip 2.

resulting from its delayed nodulation. The inhibitory activityof nodulation in side B correlated with the number ofnodules formed in the initially susceptible region of the rootsat the time of the first inoculation in side A.Nodule occupancy in coinoculation experiments. While both

wild-type and mutant bacteria appear to initiate infections atthe same time (Fig. 6), nodules induced by the mutant strainemerged later than those induced by the wild type (Fig. 5).This suggests that mutant Rm6963 was probably affected inits ability to develop infections at a normal rate rather than inits timing in initiating infections. Symbiotic deficiencies ofRm6963 were clearly reflected in its altered ability to com-pete for nodule occupancy. A preliminary experiment inplastic pouches with 106 bacteria of each strain per plant inthe inoculum revealed that Rm2011 occupied all analyzeduppermost nodules on the primary root. In similar experi-ments (but using plants grown in pots), we examined noduleoccupancy within a range of mutant-to-wild-type-bacteriaratios (Fig. 7). When mutant and wild-type inocula wereequivalent in number, only Rm2011 was found in the upper-most nodules examined. One hundred times more mutantbacteria was necessary to produce a 1:1 nodule occupancyratio (Fig. 7A). Analysis of secondary-root nodules gavesimilar results (Fig. 7B). Competitiveness was estimatedquantitatively by using a previously described mathematicalmodel (2) (Fig. 8). Double-log representation of the ratio ofthe cells of each strain in nodules (P) to those in the inocula(I) consistently fitted the experimental data (r = 0.99). Thecompetitive index C2011:6963 (for strain Rm2011 to Rm6963)had a positive value of 1.52, indicating that if I2011 = I6963, anodule occupancy ratio of P2011 to P6963 near 30:1 shouldoccur. On the other hand, to reach a value of P2011/P6963 = 1,an 12011-to-I6963 inoculum ratio of 1:100 had to be used.

DISCUSSIONWe examined the role of LPS during the symbiotic asso-

ciation ofR meliloti with alfalfa. Following random trans-poson TnS mutagenesis of wild-type bacteria, mutantRm6963 was isolated because of its rough colony morphol-ogy on TY agar medium. The abnormal sensitivity to growthin the presence of detergents suggested initially that themutant could be affected in its LPS. A careful analysisrevealed that the mutant was more sensitive to the hydro-phobic compounds DOC and SDS. Transduction of the Nmr

100

80

60

40

z

IDC)C)0

D00z

20

o1 E

100

80

60

40

20

--9--o 0~o-oO O

A. Primary root

0

00

0. 0'S l 0. U U U U[-1O1, 10 100 100 1E 1E51oE6 lEE:-1 1 10 io100 l0 14 l5 16 1

O1-1E-1

:7

RATIO OF 6963 TO 2011 IN INOCULUM( 6963 : 2011 )

FIG. 7. Competition experiments for nodulation of alfalfa withRm2011 and Rm6963 as inoculant over a range of bacterial ratios.The occupancy of nodules by each strain in primary roots (A) andlateral roots (B) was determined by using DOC and neomycin asresistance markers of strains Rm2011 (-) and Rm6963 (0), respec-tively (see Materials and Methods). The intersections of both curvesin the graphs indicate the ratios (on the abscissas) of bacteria withwhich a 50% occupancy was obtained. Bacterial growth in bothmedia (0) was considered an indication of cooccupancy or partialreversion of the mutant. This was observed only in secondary-rootnodules and occurred in less than 10% of the number of nodulesexamined. 6963:2011, Rm6963 to Rm2011.

J. BACTERIOL.

R MELILOTI LPS MUTANT ALTERED IN SYMBIOSIS

2.0

I-IC\)

Q0

l l0

0

1.5

1.0

0.5

0.0

-0.5

-1.0

-1.5

-2.0 . ,-5.0 -4.0 -3.0 -2.0 -1.0 0.0

log L12011/169631FIG. 8. Graphic estimation of the relative competitiveness of strains Rm2011 and Rm6963. A function of the nodule occupancies of the

strains was regressed as a function of the inoculum ratio as previously published (2). The ratio of the proportion of nodules in the primaryroot occupied by each strain (P2011:P6963) was logarithm transformed and was regressed on the logarithm of the ratio of the two strains in theinoculum (I2011:6963) as follows: log (P2011/P6963) = C2011:C6963 + k log(I2011/I6963). The competitiveness index (C2011:C6963) had a positivevalue, indicating that strain Rm2011 occupies a larger proportion of the nodules than strain Rm6963 when I201, = I6963. A ratio of P20ll/P6963= 1 was reached with an inoculum ratio of I201, to I6963 near 1:100.

marker proved that this mutant phenotype was caused by thetransposon insertion.

Gel permeation of hot phenol-water extracts in SepharoseCL413 separated three fractions in both strains and clearlyshowed that Rm6963 in fact carried an altered LPS. How-ever, although LPS eluted with the first two fractions in boththe wild-type and the mutant strain, extracts from Rm6963showed an increased proportion of KDO-containing materialnear the void volume (PI). The appearance of this KDO-positive material strongly suggests that mutant LPS couldform larger aggregates than wild-type molecules, probablybecause of a shorter LPS saccharide region. In other gram-negative bacteria, LPS with a short saccharide moietyformed large macromolecular aggregates in solution (30).Electrophoretic mobilities confirmed that LPS componentsfrom Rm6963 had lower molecular weights than homologouscomponents in the wild type. The fact that by SDS-PAGE,component LPS II (LPS Ila in the mutant) was present inboth PI and PII could derive from aggregates of this compo-nent ranging broadly in size and containing common molec-ular species.

Results from ELISA and immunoblotting assays showedthat LPS structures recognized by the homologous serum inthe wild-type LPS were not detected in the mutant. Since theimmunoreactivity was monitored with Igs from a polyclonalantiserum, we do not know whether the TnS insertionaffected several epitopes simultaneously or a restrictednumber of highly immunodominant structures. Electropho-retic analysis of LPS II and LPS IIa showed that the Tn5insertion altered both gel migration and immunoreactivity ofthis component in the mutant. If no pleiotropic effects arepresent, the changes that result in the shift from LPS II toLPS Ila may be the same ones that cause the shift from LPS

I to LPS Ia in the complete LPS forms. However, mutantLPS molecules can be altered in several regions. Formally,the total loss of immunoreactivity in the mutant LPS can bedue to alterations in (i) epitopes common to LPS I and LPSII (i.e., core epitopes) or (ii) epitopes common to LPS I andLPS II and epitopes exclusive to LPS I. Alteration ofepitopes present only in LPS I should have preserved thereactivity of LPS Ila. Our results show that a great propor-tion of the LPS immunoreaction is supported by LPS II.Surprisingly, no residual immunoreaction against the re-maining LPS Ia in the mutant was detected. LPS Ia wasexpected to harbor a reactive 0 antigen, as in other gram-negative bacteria. These results suggest that in fact theremaining LPS groups unaltered by the TnS mutation areweak antigenically in the wild type. On the other hand, thestrong antigenic character of LPS II contrasts with thebehavior of LPS from S-type bacteria, in which antibodiesare raised mainly against the 0 antigen. The immunoreactionofR meliloti LPS resembles in a sense the immunochemis-try of R-type mutants from members of the family Entero-bacteniaceae. These mutants and R. meliloti have LPS withhigh KDO/hexose ratios (12). In R meliloti, however, weobserved a marked antigenicity of LPS II occurring even inthe presence of the larger LPS I forms.

In studying the role of LPS in symbiosis, we found thatnodules formed by strain Rm6963 emerged later than thoseformed by the parent. However, both strains induced nod-ules to appear in the same locations along the primary rootsand to emerge at comparable rates. This suggests that themutation in strain Rm6963 did not delay the responsivenessof the host to the infecting bacteria or reduce the speed withwhich new infections occurred. The delay in nodule emer-gence suggests that although mutant bacteria were able to

5949VOL. 174, 1992

5950 LAGARES ET AL.

reach the infectible sites at the right time, it took anadditional amount of time for the infection sites to becomeactive or for the actual infections to start developing. Inexperiments in which nodulation of the mutant was lowerthan that of the wild type, a lower number of infections mayhave occurred. However, the timing at which infectionsfollowed root development was unaltered and paralleled thewild-type behavior.The impaired-nodulation phenotype of Rm6963 appears to

compromise its ability to compete for nodulation with wild-type bacteria. Inoculum ratios favoring the mutant by 100:1were required to obtain an equal level of nodule occupancy.With equivalent numbers of competing bacteria, wild-typeR meliloti infected the roots as if nodulation occurredvirtually in the absence of the mutant. In these circum-stances, the wild type had the symbiotic advantage ofevoking first the autoregulatory response that controls nod-ulation (7) and of suppressing infections that were initiatedlater in root development as well as those that were alreadyinitiated but developed slowly.We propose that the mutation in strain Rm6963 has altered

the relationship between the time of establishment of firstinfections and the rate of appearance of autoregulation.Single-inoculation experiments and competition experimentssuggest that nodules formed by Rm6963 in primary roots aresubject to feedback control of nodule formation. However,the mutant bacteria were able to elicit only a defectiveautoregulatory response. The inoculation with Rm6963 onone side of a split-root system suppressed nodulation by awild-type R meliloti strain on the other side to a low extentwhen it was inoculated 24 h later. This deficiency resulted ina decrease in control of the number of lateral-root nodules.One month after inoculation, Rm6963 had formed approxi-mately five times the number of nodules produced by thewild type. This effect is probably the expression of homeo-stasis to compensate for delayed nodulation. We have notexamined whether the deficiency in autoregulatory re-sponses is due to a defect in the stimulation of cortical-celldivision, a step proposed to be relevant in nodulation control(9). It has been suggested that R. meliloti LPS could beinvolved in at least two stages during nodule formation (48).Our results suggest a role for LPS during the initial devel-opment of infections rather than during infection initiation.However, it must be considered that our evidence is based inone mutant and that it is the formal possibility that some ofthe mutant phenotypes are due to separate mutations inRm6963. Even if all phenotypes are due to the same muta-tion, the symbiotic defects could arise, at least formally,from alterations in other molecules besides LPS. Furthergenetic characterization of this mutant is being studied.

Clearly, as indicated by the genetic background ofRm6963, LPS changes in this mutant are not related to thewell-characterized lpsZ suppressor gene naturally absent inall SU-47-derivative strains (63). Another gene(s) must beinvolved. However, structures altered in Rm6963 might stillbe necessary for the expression of a chimeric LpsZ+ pheno-type in strain Rm2011. Other SU-47 LPS mutants carryingthe lpsZ gene in an exoB background were still unable to fixnitrogen (62). New correlations between known Ips genesand their functions during the interaction with the host plantshould provide insight into the role of this bacterial surfacecomponent in symbiosis.

In summary, current evidence suggests that R melilotiLPS is indeed involved in the symbiotic interaction withalfalfa by (i) providing a structure able to recognize the plantroot during bacterial adsorption (41), (ii) modulating strain

competitiveness during infection (the present work), and (iii)supporting a compatible association with the host plantindependently or in conjunction with exopolysaccharide (48,62, 63). The goal is now to determine the molecular mecha-nism(s) of LPS action.

ACKNOWLEDGMENTSWe are greatly indebted to C. A. Fossati for invaluable help in

setting up the immunotechnics, and we thank A. Campana forexcellent technical assistance.A.L. has been awarded fellowships from CONICET (Argentina)

and from the Alexander von Humboldt Foundation AvH (Germa-ny). G.F. is a member of the Research Career of CONICET(Argentina). This work was supported by grants from SECYT,CONICET, and CICBA (all from Argentina), SAREC CooperationProject (from Sweden), and Stiftung Volkswagenwerk (I/64838[from Germany]). Page charges for this article were paid by AvH.

REFERENCES1. Bauer, W. D. 1981. Infection of legumes by rhizobia. Annu.

Rev. Plant Physiol. 32:407-449.2. Beattie, G. A., M. K. Clayton, and J. Handelsman. 1989.

Quantitative comparison of the laboratory and field competitive-ness of Rhizobium leguminosarum biovar phaseoli. Appl. En-viron. Microbiol. 55:2755-2761.

3. Beringer, J. E. 1974. R factor transfer in Rhizobium leguminosa-rum. J. Gen. Microbiol. 84:188-198.

4. Bhuvaneswari, T. V., K. K. Mills, D. K. Christ, M. R. Evans,and W. D. Bauer. 1983. Effects of culture age on symbioticinfectivity of Rhizobium japonicum. J. Bacteriol. 153:443-451.

5. Borthakur, D., C. E. Barber, J. W. Lamb, M. J. Daniels, J. A.Downie, and A. W. B. Johnston. 1986. A mutation that blocksexopolysaccharide synthesis prevents nodulation of peas byRhizobium leguminosarum but not of beans by Rhizobiumphaseoli and is corrected by cloned DNA from Rhizobium or thephytopathogen Xanthomonas. Mol. Gen. Genet. 203:320-323.

6. Brink, B. B., J. Miller, R. W. Carlson, and K. D. Noel. 1990.Expression of Rhizobium leguminosarum CFN42 genes forlipopolysaccharide in strains derived from different R legumi-nosarum soil isolates. J. Bacteriol. 172:548-555.

7. Caetano-Anoll6s, G., and W. D. Bauer. 1988. Feedback regula-tion of nodule formation in alfalfa. Planta 175:546-557.

8. Caetano-Anolles, G., and G. Favelukes. 1986. Quantitation ofadsorption of rhizobia in low numbers to small legume roots.Appl. Environ. Microbiol. 52:371-376.

9. Caetano-Anoll6s, G., and P. M. Gresshoff. 1990. Early inductionof feedback regulatory responses governing nodulation in soy-bean. Plant Sci. 71:69-81.

10. Caetano Anolles, G., A. Lagares, and W. D. Bauer. 1990.Rhizobium meliloti exopolysaccharide mutants elicit feedbackregulation of nodule formation in alfalfa. Plant Physiol. 91:368-374.

11. Carlson, R. W. 1982. Surface chemistry, p. 199-204. In W. J.Broughton (ed), Nitrogen fixation, vol. 2, Rhizobium. Claren-don Press, Oxford.

12. Carlson, R. W. 1984. The heterogeneity of Rhizobium lipopoly-saccharides. J. Bacteriol. 158:1012-1017.

13. Carlson, R. W., S. Kalembasa, D. Turowski, P. Pachori, andK. D. Noel. 1987. Characterization of the lipopolysaccharidefrom a Rhizobium phaseoli mutant that is defective in infectionthread development. J. Bacteriol. 158:1012-1017.

14. Carlson, R. W., U. Ramadas Bhat, and B. Reuhs. 1991. Rhizo-bium lipopolysaccharides: their structures and evidence fortheir importance in the nitrogen-fixing symbiotic infection oftheir host legumes, p. 39-50. In P. M. Gresshoff (ed), Currenttopics in plant molecular biology, vol. 1. Chapman and Hall,New York.

15. Carlson, R. W., R. E. Sanders, C. Napoli, and P. Albersheim.1978. Host-symbiont interaction. III. Purification and partialcharacterization of Rhizobium lipopolysaccharides. Plant Phys-iol. 62:912-917.

16. Carlson, R. W., R Shatters, J.-L. Duh, E. Turnbull, B. Hanley,

J. BAcTERIOL.

VMELILOTI LPS MUTANT ALTERED IN SYMBIOSIS 5951

B. G. Rolfe, and M. A. Djordjevic. 1987. The isolation andpartial characterization of the lipopolysaccharides from severalRhizobium trifolii mutants affected in root hair infection. PlantPhysiol. 84:421-427.

17. Carlson, R. W., and M. Yadav. 1985. Isolation and partialcharacterization of the extracellular polysaccharides from fast-growing Rhizobiumjaponicum USDA 205 and its Nod- mutant,HC205, which lacks the symbiotic plasmid. Appl. Environ.Microbiol. 50:1219-1224.

18. Carrion, M., U. Ramadas Bhat, B. Reuhs, and R. W. Carlson.1990. Isolation and characterization of the lipopolysaccharidefrom Bradyrhizobium japonicum. J. Bacteriol. 172:1725-1731.

19. Cava, J. R., P. M. Elias, D. A. Turowski, and K. D. Noel. 1989.Rhizobium leguminosarum CFN42 genetic regions encodinglipopolysaccharide structures essential for complete noduledevelopment on bean plants. J. Bacteriol. 171:8-15.

20. Chacravarty, A. K, W. Zurkowsky, J. Shine, and B. G. Rolfe.1982. Symbiotic nitrogen fixation: molecular cloning of Rhizo-bium genes involved in exopolysaccharide synthesis and effec-tive nodulation. J. Mol. Appl. Genet. 1:585-596.

21. Chen, H., M. Batley, J. M. Redmond, and B. G. Rolfe. 1985.Alteration of the effective nodulation properties of a fast-growing broad host range Rhizobium due to changes in ex-opolysaccharide synthesis. J. Plant Physiol. 120:331-349.

22. Clark, M. F., and H. N. Adams. 1977. Characteristics of themicroplate method of enzyme-linked immunosorbent assay forthe detection of plant viruses. J. Gen. Virol. 34:475-483.

23. Clover, R. H., J. Kieber, and E. R. Signer. 1989. Lipopolysac-charide mutants of Rhizobium meliloti are not defective insymbiosis. J. Bacteriol. 171:3961-3967.

24. Dazzo, F. B., G. L. Truchet, R I. Hollingsworth, E. M. Hrabak,H. Stuart Pankratz, S. Philip-Hollingsworth, J. L. Salzwedel, K.Chapaman, L. Appenzeller, A. Squartini, D. Gerhold, and G.Orgambide. 1991. Rhizobium lipopolysaccharide modulates in-fection thread development in white clover root hairs. J. Bac-teriol. 173:5371-5384.

25. DeMaagd, R. A., A. S. Rao, I. H. M. Mulders, L. Goosen-deRoo, M. C. M. Van Loosdrecht, C. A. Wjifelman, and B. J. J.Lugtenberg. 1989. Isolation and characterization of mutants ofRhizobium leguminosarum bv. viciae 248 with altered lipopolry-saccharides: possible role of surface charge or hydrophobicityin bacterial release from the infection thread. J. Bacteriol.171:1143-1150.

26. D6narie, J., and P. Roche. 1992. Rhizobium nodulation signals,p. 295-324. In D. P. S. Verma (ed.), Molecular signals inplant-microbe interactions. CRC Press, Boca Raton, Fla.

27. Dubois, M., K. A. Gilles, J. K. Hamilton, P. A. Rebers, and F.Smith. 1956. Colorimetric method for determination of sugarsand related substances. Anal. Chem. 28:350.

28. Fahraeus, G. 1957. The infection of clover root hairs by nodulebacteria studied by a simple glass slide technique. J. Gen.Microbiol. 16:374-381.

29. Finan, T. M., E. Hartwieg, K. LeMieux, K Bergman, G. C.Walker, and E. Signer. 1984. General transduction in Rhizobiummeliloti. J. Bacteriol. 159:120-124.

30. Galanos, C., and 0. Lfideritz. 1975. Electrodialysis of lipopoly-saccharide and their conversion to uniform salt forms. Eur. J.Biochem. 54:603-610.

31. Geremia, R. A., S. Cavaignac, A. Zorreguieta, N. Toro, J.Olivares, and R. A. Ugalde. 1987. A Rhizobium meliloti mutantthat forms ineffective pseudonodules in alfalfa produces ex-opolysaccharide but fails to form P-(1->2) glucan. J. Bacteriol.169:880-884.

32. Goosen-de Roo, L., R. de Maagd, and B. J. J. Lugtenberg. 1991.Antigenic changes in lipopolysaccharide I ofRhizobium legumi-nosarum bv. viciae in root nodules of Vicia sativa subsp. nigraoccur during release from infection threads. J. Bacteriol. 173:3177-3183.

33. Halverson, J. L., and G. Stacey. 1986. Signal exchange inplant-microbe interactions. Microbiol. Rev. 50:193-225.

34. Jensen, H. L. 1942. Nitrogen fixation in leguminous plants. I.General characters of root nodule bacteria isolated from speciesof Medicago and Trifolium in Australia. Proc. Linn. Soc.

N. S. W. 66:98-108.35. Kamberger, W. 1979. Ouchterlony double diffusion study on the

interaction between legume lectins and rhizobial cell surfaceantigens. Arch. Microbiol. 121:83-90.

36. Kannenberg, E. L., S. S. Sindhu, and N. J. Brewin. 1988.Correlation between Rhizobium lipopolysaccharide structureand the ability to grow under calcium stress, p. 476. In H.Bothe, F. J. de Brujin, and W. E. Newton (ed), Nitrogenfixation: hundred years after. Gustav Fischer, Stuttgart, Ger-many.

37. Karkhanis, Y. D., J. Y. ZeItner, J. J. Jackson, and D. J. Carlo.1978. A new and improved microassay to determine 2-keto-3-deoxyoctonate in lipopolysaccharide of gram-negative bacteria.Anal. Biochem. 85:595-601.

38. Kato, G., M. Murayama, and M. Nakamura. 1979. Role oflectins and lipopolysaccharides in the recognition process ofspecific legume-Rhizobium symbiosis. Agric. Biol. Chem. 43:1085-1092.

39. Kropinsky, A. M., D. Berry, and E. P. Greenberg. 1986. Thebasis of silver staining of bacterial lipopolysaccharides in poly-acrylamide gels. Curr. Microbiol. 13:29-31.

40. Laemmli, U. K 1970. Cleavage of structural proteins during theassembly of the head of bacteriophage T4. Nature (London)227:680-685.

41. Lagares, A., and G. Favelukes. 1990. Rhizobium meliloti saccha-ride moiety involved in early symbiotic recognition by alfalfa, p.265. In P. M. Gresshoff, L. Evans Roth, G. Stacey, and W. E.Newton (ed.), Nitrogen fixation: achievements and objectives.Chapman and Hall, New York.

42. Leigh, J. A., E. R. Signer, and G. C. Walker. 1985. Exopolysac-charide deficient mutants that form ineffective nodules. Proc.Natl. Acad. Sci. USA 82:6231-6235.

43. Long, S. R. 1989. Rhizobium-legume nodulation: life together inthe underground. Cell 56:203-214.

44. Long, S. R. 1989. Rhizobium genetics. Annu. Rev. Genet.23:483-506.

45. Muller, P., M. Hynes, D. Kapp, K. Niehaus, and A. Puihler. 1988.T'wo classes of Rhizobium meliloti infection mutant differ inexopolysaccharide production and in coinoculation propertieswith nodulation mutants. Mol. Gen. Genet. 211:17-26.

46. Noel, K. D., K. A. Vandenbosch, and B. Kulpaca. 1986. Muta-tions in Rhizobiumphaseoli that lead to arrested development ofinfection threads. J. Bacteriol. 168:1392-1401.

47. Priefer, U. B. 1989. Genes involved in lipopolysaccharide pro-duction and symbiosis are clustered on the chromosome ofRhizobium leguminosarum biovar viciae VF39. J. Bacteriol.171:6161-6168.

48. Putnoky, P., G. Petrovics, A. Kereszt, E. Grosskopf, D. T. C.Ha, Z. Banfalvi, and A. Kondorosi. 1990. Rhizobium melilotilipopolysaccharide and exopolysaccharide can have the samefunction in the plant-bacterium interaction. J. Bacteriol. 172:5450-5458.

49. Rolfe, B. G., and P. M. Gresshoff. 1988. Genetic analysis oflegume nodule initiation. Annu. Rev. Plant Physiol. Plant Mol.Biol. 39:297-319.

50. Russa, R., T. Urbanik, E. Kowalczuk, and Z. Lorkiewicz. 1982.Correlation between the occurrence of plasmid pUCS202 andlipopolysaccharide alterations in Rhizobium. FEMS Microbiol.Lett. 13:161-165.

51. Russa, R., T. Urbanik, W. Zurkowski, and Z. Lorkiewicz. 1981.Neutral sugars in lipopolysaccharides of Rhizobium tifolii andits non-nodulating mutant. Plant Soil 161:81-85.

52. Sanderson, K. E., T. MacAlister, and J. W. Costerton. 1974.Lipopolysaccharide-deficient (rough) mutants of Salmonella ty-phimurium to antibiotics, lysozyme and other agents. Can. J.Microbiol. 20:1135-1145.

53. Simon, R., U. Priefer, and A. Puihler. 1983. A broad host rangemobilization system for in vivo engineering: transposon muta-genesis in gram negative bacteria. Bio/Technology 1:784-791.

54. Sindhu, S. S., N. J. Brewin, and E. L. Kannenberg. 1990.Immunochemical analysis of lipopolysaccharides from free-living and endosymbiotic forms of Rhizobium leguminosarum.J. Bacteriol. 172:1804-1813.

VOL. 174, 1992

5952 LAGARES ET AL.

55. Somasegaran, P., and H. J. Hoben. 1985. Methods in legume-Rhizobium technology, p. 77-79. NifrAL-MIRCEN.

56. Stacey, G., J.-S. So, L. Evans Roth, S. K. Bhagya Lakashmi, andR. W. Carlson. 1991. A lipopolysaccharide mutant of Brady-rhizobium japonicum that uncouples plant from bacterial differ-entiation. Mol. Plant-Microbe Interact. 4:332-340.

57. Strum, S., P. Fortnagel, and K. N. Timmis. 1984. Immunoblot-ting procedure for the analysis of electrophoretically-fraction-ated bacterial lipopolysaccharide. Arch. Microbiol. 140:198-201.

58. Tao, H., N. J. Brewin, and K. D. Noel. 1990. Rhizobiumleguminosarum CFN42 lipopolysaccharide antigenic changesinduced by environmental conditions. J. Bacteriol. 174:2222-2229.

59. Tsai, C. M., and C. E. Frasch. 1982. A sensitive silver stain fordetecting lipopolysaccharides in polyacrylamide gels. Anal.

Biochem. 119:115-119.60. Vincent, J. M. 1979. A manual for the practical study of root

nodule bacteria. Handbook no. 15 C. IBP, Oxford.61. Westphal, O., and K. Jann. 1965. Bacterial lipopolysaccharide.

Methods Carbohydr. Chem. 5:83-91.62. Williams, M. N. V., R. I. Hollingsworth, P. M. Brzoska, and

E. R. Signer. 1990. Rhizobium meliloti chromosomal loci re-quired for suppression of exopolysaccharide mutations by lipo-polysaccharide. J. Bacteriol. 172:6596-6598.

63. Williams, M. N. V., R. I. Hollingsworth, S. Klein, and E. Signer.1990. The symbiotic defect of Rhizobium meliloti exopolysac-charide mutants is suppressed by lpsZ+, a gene involved inlipopolysaccharide biosynthesis. J. Bacteriol. 172:2622-2632.

64. Williamson, P., K. Mattocks, and R. A. Schlegel. 1983. Merocy-anine 540, a fluorescent probe sensitive to lipid packing. Bio-chim. Biophys. Acta 732:387-393.

J. BACTERIOL.